Installing into a new cluster

The full install reference — sizing modes, ingress, autoscaling, and monitoring.

If you just want telark running, the Quickstart is shorter. This page is the reference: what the one command actually installs, and every flag worth knowing before you run it in production.

The command

helm install telark oci://ghcr.io/telark/charts/telark -n telark --create-namespace

One release installs the services, the custom resource definitions, the bundled policy engine, the coordination and messaging layers, and the dashboard. The CRDs are a bundled subchart — there is no separate CRD step and no bootstrap chart to run first. They carry helm.sh/resource-policy: keep, so they survive helm uninstall.

Managing CRDs out of band? --set crds.enabled=false.

Everything is configured with --set key=value on the same command line. Helm does not remember those flags across upgrades — re-pass them on every helm upgrade.

Sizing modes

app.mode sizes every telark service from a single flag: replicas, resource requests, client rate limits, and PodDisruptionBudgets.

Mode	Replicas	What it changes
`minimal`	1	Dev, demos, evaluation. No disruption budgets. Smallest resource requests.
`standard` (default)	2	Small-to-mid production. Disruption budgets on.
`performance`	3	Large clusters. Autoscaling on automatically (min 3, max 5). Requires a ReadWriteMany storage class for the exporter.

helm install telark oci://ghcr.io/telark/charts/telark -n telark --create-namespace \
  --set app.mode=performance \
  --set app.persistence.storageClass=<rwx-class>

app.mode sizes telark's own services only. Helm resolves a subchart's values before the mode is known, so the bundled dependencies — the policy engine, the coordination layer, the event stream, metrics — ship fixed production-grade defaults owned by the chart, identical in every mode. There is nothing to tune there.

App flags

Flag	Default	Description
`app.mode`	`standard`	Sizing mode (above).
`app.name`	`telark`	Identity prefix for every resource name and API group.
`app.namespace`	`telark`	Install namespace.
`app.image.registry`	`telark`	Registry / org hosting the service images.
`app.image.pullPolicy`	`Always`	Image pull policy.
`app.image.pullSecrets`	`[]`	Pull secrets, if you mirror the images privately.
`app.persistence.size`	`10Gi`	Exporter snapshot PVC size.
`app.persistence.storageClass`	`""`	`""` uses the cluster default. `performance` needs a ReadWriteMany class.
`app.persistence.accessMode`	`ReadWriteOnce`	Exporter PVC access mode.
`app.crdGuard.enabled`	`false`	Admission guard: only the owning service accounts may write telark custom resources.
`app.crdGuard.enforce`	`false`	With the guard on, `false` audits and `true` rejects.
`app.auth.bootstrap.admins[0]`	`contact@telark.io`	Emails granted Admin on first login. Indexed: `[0]`, `[1]`, …
`app.auth.passkey.selfRegistration`	`"true"`	`"false"` blocks new passkey registration; you then need a bootstrap admin.

Ingress (optional, off by default)

Port-forwarding is enough to evaluate telark. For a real hostname and TLS, turn the Ingress on — you need an ingress controller already in the cluster.

helm install telark oci://ghcr.io/telark/charts/telark -n telark --create-namespace \
  --set ingress.enabled=true \
  --set ingress.className=nginx \
  --set ingress.host=telark.example.com

Key	Default	Description
`ingress.enabled`	`false`	Create an Ingress for the dashboard.
`ingress.className`	`""`	IngressClass, e.g. `nginx`.
`ingress.host`	`""`	Hostname. `""` matches any host.
`ingress.service`	`ui`	Which `services.<key>` to route to.
`ingress.path` / `ingress.pathType`	`/` / `Prefix`	Route path and match type.
`ingress.tls`	`[]`	TLS blocks, e.g. `[{secretName: telark-tls, hosts: [telark.example.com]}]`.
`ingress.annotations`	`{}`	Controller annotations — cert-manager issuers and so on.

Autoscaling

The stateless services — auth, discovery, enrichment, notifier, ui — can run behind a HorizontalPodAutoscaler (autoscaling/v2, CPU-based). The exporter never autoscales: it holds a ReadWriteOnce volume.

performance mode turns autoscaling on automatically (min 3, max 5). In any mode you can enable or tune it per service:

helm install telark oci://ghcr.io/telark/charts/telark -n telark --create-namespace \
  --set services.discovery.autoscaling.enabled=true \
  --set services.discovery.autoscaling.minReplicas=2 \
  --set services.discovery.autoscaling.maxReplicas=8 \
  --set services.discovery.autoscaling.targetCPUUtilizationPercentage=70

Key	Default	Description
`services.<svc>.autoscaling.enabled`	`false`	Turn the HPA on for that service (on in `performance`).
`services.<svc>.autoscaling.minReplicas`	`1`	Replica floor.
`services.<svc>.autoscaling.maxReplicas`	`3`	Replica ceiling.
`services.<svc>.autoscaling.targetCPUUtilizationPercentage`	`80`	Scale-up CPU target.
`services.<svc>.autoscaling.targetMemoryUtilizationPercentage`	(unset)	Optional memory target.

Set the same keys under app.serviceDefaults.autoscaling to change the default for every service at once. When a service autoscales, Helm stops managing its replica count so the HPA and Helm do not fight over it.

Monitoring (opt-in)

telark can emit a ServiceMonitor that scrapes every service's /metrics. It is off by default because it needs the Prometheus Operator CRDs already in the cluster.

helm install telark oci://ghcr.io/telark/charts/telark -n telark --create-namespace \
  --set monitoring.serviceMonitor.enabled=true \
  --set monitoring.serviceMonitor.labels.release=<prometheus-release>

The label is the part people miss: Prometheus only picks up a ServiceMonitor whose labels match its serviceMonitorSelector — usually release: <your-release>. Omit it and the monitor is silently ignored.

Key	Default	Description
`monitoring.serviceMonitor.enabled`	`false`	Create the ServiceMonitor.
`monitoring.serviceMonitor.labels`	`{}`	Labels matching Prometheus's `serviceMonitorSelector`.
`monitoring.serviceMonitor.path`	`/metrics`	Scrape path.
`monitoring.serviceMonitor.interval`	`30s`	Scrape interval.

Upgrading

helm upgrade telark oci://ghcr.io/telark/charts/telark -n telark \
  --set app.mode=<mode>

Re-pass every --set and -f flag you used at install. See Operations → Upgrading.

Every value

The complete field list — every telark value and every pinned dependency value — is the chart's auto-generated charts/telark/VALUES.md in the repository.

Installing into a new cluster

On this page